When it comes to Prime Minister, most Canadians would like to see Jack Layton get the post according to the latest Ipsos-Reid poll done for Global National.

Forty-five per of Canadians said they think Jack Layton would make the best prime minister. Forty-two per cent said Stephen Harper and 13 per cent choose Michael Ignatieff. (Click here for the full poll and methodology)

That doesn’t mean that Layton will become prime minister however. The final results will only be in on May 2, when Canadians officially cast their ballots.

Until then, the country will rely on polls to monitor the progress of the election. But what are polls and what should Canadians consider when using them?

What is a poll and why is it used?

A poll isn’t a simple popularity contest. It is a way to gauge the opinion of a population, by asking targeted questions of a sample. The responses are then extrapolated to represent the general population.

Polls are used for a simple reason – to determine who could win, says Paul Adams, an associate professor at Carleton’s School of Journalism and former pollster at EKOS.

“At the end of the day, who wins is the fundamental question that an election is posing,” says Adams. “Maybe you don’t want to know the score in the Canucks game until it is all over, but a lot of us like to see it as the goals come in. We find it interesting to know how it progresses.”

“The polls give us information albeit partial and uncertain about the status of the various different parties,” he said. “That can inform people’s decisions about their vote.”

What to look for when reading a poll?

Elections Canada requires the publishers of public opinion surveys during elections to publish some facts about the methodology, so readers can gauge how reliable the poll is.

Anyone transmitting the results of a poll has to include the name of the sponsor and the company that did the poll, which will help readers determine if the poll is objective.

Readers should also have access to the date when the poll was conducted and the size of the sample that was consulted to test reliability.

Finally, organizations have to disclose the margin of error, one of the most important pieces of information, according to Robert MacDermid, a political science professor at York University.

If a poll says Jack Layton has 20 per cent of Canadians supporting him, but there is a margin of error of +/- 3 per cent, that means the support is actually between 23 and 17 per cent, he explained.

One thing most polls won’t tell a voter is how the race is unfolding locally, which limits its effectiveness for strategic voters.

Why do polls have different results?

Global’s last poll done by Ipsos-Reid showed that the NDP surged ahead of the Liberals with 24 per cent of the vote, up from 19 per cent on April 7. (Click here to view the poll and methodology)

Just a week ago, Canadians had different opinions on who should get the top job as well, according to Ipsos-Reid. The numbers were 47 per cent for Harper, 35 per cent for Layton and 19 per cent for Ignatieff.

Differences can reflect legitimate changes in public opinion or variations in methodology, says Adams.

Reading polls is getting trickier because polling companies are relying more on online respondents instead of phone surveys, he said.

Adams said online polling doesn’t have the same breadth of sampling and that it relies on people signing up to do the poll, meaning they are not random.

“I’m not saying that no online poll is ever accurate, but in the past what we have seen is online polls are much less reliable election after election than phone polls,” he said.

Angus Reid, CEO of Vision Critical and Angus Reid Public Opinion disagrees with that assessment of online polls.

“We’ve covered nine election sin Canada since we started online polling in 2007 and in only one of those elections, Alberta 2008, we did underestimate the size of the Ed Stelmach majority, but more recently, we have been accurate,” he said, providing the numbers to verify his claims.

Instead of comparing phone to online polling, readers need to consider the method used online, Reid says.

Many polling companies, like Angus Reid, spend years building up their online forums and in doing so work hard to make it representative of the Canadian population, demography and geography.

Adams said what voters should look for is what the polls have in common over the course of the election campaign.

This time around, the polls are all showing the Conservatives staying stable, the NDP on the rise and the Liberals on a downward trend, he said.

As the campaign goes on, most of the polls converge because people have made up their minds, which minimize the impact of methodology.

Polls that have results that are way off of the norm are usually discarded by polling companies as rogue.

Alternative gauges of progress

Polls aren’t the only way to gauge progress, according to MacDermid, who uses Layton’s recent rise in Quebec as an example.

“There are all sorts of evidence that people, especially in Quebec, are considering voting for Jack,” he said. “You’d report these other things; that he is spending all his time there; that 1000 people showed up at a rally; and that other political parties are attacking him.”

And polls aren’t the only way to determine how you want to vote, MacDermid says. Voters should consider the party platforms and what the leaders say about the issues that matter to the individual voter.

“If you don’t want to take (polls) into consideration then don’t,” says Adams. “No one is ever going to criticize you for saying I voted for X because I thought that party had the best policies and the best leader and its views fit with my views.”